Regarding the precision of combined domains including Jacobs and Langen's Sharing there is a core of techniques, such as the standard integration with freeness and linearity information, that are widely used and well accepted. However, a number of other proposals for refined domain combinations have been circulating more or less clandestinely for years. One feature that is common to these proposals is that they do not seem to have undergone experimental evaluation. We question whether significantly more precision is obtainable thanks to these techniques. In particular, we discuss and/or experimentally evaluate: helping Sharing with definitely ground variables computed with Pos; the incorporation of explicit structural information into the domain of analysis; more sophisticated ways of integrating Sharing and Pos; the issue ering the bindings in the computation of the abstract mgu; an original proposal concerning the addition of a domain recording the set of variables that are dee...
Roberto Bagnara, Enea Zaffanella, Patricia M. Hill